A. Melander, G. Sismanidis and D. Grenouilleau (European Commission, Directorate General for Economic and Financial Affairs)
This paper has updated the assessment of the Commission’s forecasts’ track record from 1999 by extending the observation period from 1969-1997 to also take into account the forecasts and outcome for the years 1998-2005. This update has also included some further tests on e.g. informational efficiency and undertaken a comparison with the forecasts of other international institutions and those of market participants. The tests were carried out on the forecasts for real GDP growth, total investment, inflation, the unemployment rate, the general government balance and the current account to GDP ratio. Data have been processed in a broadly similar manner compared to the study of 1999 to ensure comparability to the greatest degree possible.
Overall, the Commission's forecasts continue to dispose a reasonable track record. For instance, the forecast error for the GDP forecast, as measured by the mean absolute error, has improved by 0.03 percentage point (pp.) to 0.5 pp. for the current-year outlook and by 0.08 pp. to 0.86 pp. for the year ahead. This implies that the Commission's forecasts for GDP growth has, on average, proven to be 0.5 pp. too high / low for the current year. Forecasts for the EU generally seem to be unbiased, efficient and display a high success rate for directional accuracy. The same holds true for the outlook for most Member States, although there are individual examples to the contrary. Moreover, in view of the importance of the international environment in explaining past forecast errors, it is reassuring to note that the forecasts for the largest non-EU countries generally seem to perform well. Finally, the Commission's forecasts' track record for GDP is broadly comparable with the ones of Consensus, the IMF and the OECD.
|ISBN 978-92-79-04644-5 (online)|
|ISSN 1016-8060 (print)|